Quote:
Originally Posted by GriffinRU
I will disagree with you, Patrick, about sensored motors, and there is no "Hype" here. For given application there is a given motor.
|
I'm going to disagree with you here. Sensored motors are marketed with a lot of HYPE -- The makes of sensored systems have been BSing the public for a couple years now, telling them that sensorless is inferior, in the hopes that people will buy into their hype. And some people have bought into the hype.
Try this: Take a sensored motor and measure how much phase imbalance the sensors have -- you'll be astonished and shocked at how BAD the sensors are. Typical sensor phase imbalance is around 10 degrees from sensor to sensor, and that's just PHASE imbalance. Even on high-end military motors there is at least 5 degrees of phase imbalance -- just because Hall sensors trigger points aren't precise. And timing itself is
completely ambiguous with a sensored system - - and because it is ambiguous, it can't add any meaningful data to the rotor position estimate.
Not only that, but sensors also trigger incorrectly all the time during run (by winding flux) -- and the sensor controllers blindly follow the incorrect sensor outputs...
Granted, sensored systems
can produce more torque at stall than sensorless systems. But in RC applications we can generate more torque at startup than is required using only sensorless algorithms.
You state:
Quote:
Originally Posted by GriffinRU
Timing with sensors can me more advanced and flexible if you write software accordingly
|
But that really isn't true. Once the sw PLL is locked (within a few commutation cycles) we can control timing to within .1 degrees -- how can an ambiguous sensor system with jitter and phase imbalance add to that? Adding sensors into the equations just makes the position data LESS precise.
Believe me, we build hybrid controllers (that start sensored and run sensorless) for both the military and industrial applications, and have been building those types of controllers for years.
We've done all the math, the simulations, and modeled the systems extensively. Once the sensorless software is locked, sensors add nothing but noise to the system.
Sensors are useful for generating a lot of torque at stall (> 20% PWM duty cycle lengths), but with the motors we use in RC, we can't even PWM at 20% during stall -- they would draw thousands of amps. So even at a 20% max duty at stall we can generate PLENTY of torque at startup -- Seen the videos of people backflipping Monster Trucks with the MMM from a standing start?
So there's my rant about sensors. With respect to timing and control, Sensorless = precision, Sensored = ambiguity.
Ok, so I've kinda got the "Mondays" myself. 'Cause I have to get up at 5:30am to fly to California. And I don't want to go to California.
And it's been raining here a lot, and my backyard is flooding and mulch is pouring into my pool and I keep having to spend hours cleaning it out.
And I got an E-Revo and I haven't even had time to install a Monster in it yet.
And I keep editing this post 'cause I don't want people to think I'm angry or being aggressive about this... I just want to present the facts...